Generating Sheaths For Curved Internal Edges

[Xylvangle]

Is your feature request related to a problem? Please describe. I frequently need to print parts with unavoidable curved overhangs. Print Settings > Support and Material > With sheath around support solves the issue for external edges, but not internal ones. It fails to generate supports there despite them being just as necessary. The images below demonstrate the issue with a test model. The support structure cannot be printed at the internal edge because PrusaSlicer tries to create a free-standing curved strand.

All overhang perimeters should provoke sheath generation below them.

Manually painting perimeters is a poor work-around because it requires manually panting all edges. Smart fill will not work because it still ignores internal curved overhangs. There is no way to paint on top of auto-generated supports, therefore, all overhangs, curved or not, must be manually painted.

The model in question is deliberately designed and oriented to simulate a worst case scenario. Changing its shape or orientation would obfuscate its purpose.

STL 3MF

[u89djt]

If you post the model in question, probably as the 3mf file saved from what you've screenshot, it'll be quicker to get hands on the specifics of the issue. You need to zip it before dropping into your post. (not affiliated with prusa - trawling the issues page ahead of time has saved me when I'm in a hurry later several times)

[Xylvangle]

File is attached. I didn't include it because this issue is not unique to any particular project. Any curved internal overhang will produce this result with sheaths on.

None of the supported file types in comments here are viable so here are Drive links: STL 3MF

[u89djt]

Right, and that makes sense from your perspective, but the devs need to reproduce it from their perspective, which is that they've already got 99 problems, and they need to be sure this glitch is a new one, or a helpful illustration of one they're aware of. Due diligence for a bug report includes placing the dev in exactly your situation with miminum extra work on their part. In this case, you're asking them to build an object with very general description of the form, and their version might not excite the same misbehaviour. There's a readme at the top of the issues page about good bug reporting, and a link to a really good article. You can post any file as long as it's zipped, like this: 3D Print Test 3mf.zip This is your 3mf file from Drive but stored here so there's a definite reference entitiy that they have control over. While we're here, I can confirm that I see the same behaviour too. What I haven't done is look for previous bug reports of similar phenomena.

[u89djt]

Amateur opinion: the top interface layer is there to deal with such situations; to capture the excursion of the supported layers from the path of the main support extrusions over a distance comparable with the support spacing. It can't tell the difference between a long curve and a fine sawtooth, it just knows there's stuff there in mid air. If you stop it using one of its methods for dealing with such things, it's going to fail. Perhaps this becomes a request for detecting special cases of an exactly horizontal smooth perimeter that doesn't match the local support material regime's pattern to then back off and build an exterior case and refill it with support. You'd then need to be confident that the process will terminate. Messy. Hence dense enough support and an interface layer. Workaround: When I want something to print fast and clean, I build my own supports. Here's an example for your inner curve as the Fusion 360 file, a 3mf file with your same settings, and a stand alone step file. curved overhangs.zip